FlowPipe: LLM-Enhanced Conditional Generative Flow Networks for Data Preparation Pipeline Construction
Summary: FlowPipe casts data-preparation pipeline synthesis as conditional flow generation, using trajectory balance, FiLM-modulated LLM priors, and failure-aware pruning to improve credit assignment and exploration. Across 74 datasets, it boosts accuracy 11.96% and converges 12.5× faster than SOTA baselines.
(summarized by gpt-6-luna on Oct 22 2026)
@inproceedings{ni_sigmod27,
title = {{FlowPipe: LLM-Enhanced Conditional Generative Flow Networks for Data Preparation Pipeline Construction}},
author = {Ni, Kunyu and Cao, Lei and He, Jie and Zhang, Xiaotong and Jin, Jianfeng and Dong, Junyu and Yu, Yanwei},
series = {{SIGMOD} '27},
booktitle = {Proceedings of the {ACM} {SIGMOD} International Conference on Management of Data},
publisher = {Association for Computing Machinery},
doi = {10.1145/3837112},
url = {https://dl.acm.org/doi/10.1145/3837112},
year = {2027}
}
Incoming Citations (Sorted by Pagerank)
Showing 0 of 0 citing papers.
Rank
Citing Paper
Year
Venue
Pagerank
PreviousPage 1 / 1Next
Outgoing Citations (Sorted by Pagerank)
Showing 15 of 15 cited papers.
Citations counted here include only citations to other VLDB/SIGMOD/CIDR/PODS papers in this database.